Urdu-Sindhi Speech Emotion Corpus

The Urdu-Sindhi Speech Emotion Corpus is a dataset collected at Mehran University of Engineering & Technology, Pakistan by a research team led by Dr. Zafi Sherhan Syed and Dr. Sajjad Ali Memon. The dataset consists on 1,435 audio recordings in total for seven types of emotions which include anger, disgust, happiness, neutral, sarcasm, sadness, and surprise in two low-resource languages of South Asia, that is Urdu and Sindhi.

Due to ethical restrictions we cannot release audio recordings at the moment and instead release five feature sets from the OpenSmile toolkit.

Please note that we will endeavour to compute any features for you locally and send those features back to you. If this interests you, please contact Dr. Zafi Syed at zafisherhan.shah@faculty.muet.edu.pk